求解一类新的二次规划问题的时滞投影神经网络方法  

A Delayed Projection Neural Network for Solving a New Quadratic Programming Problems

作　　者：刘德友[1] 牛九肖[1] 

机构地区：[1]燕山大学理学院,河北秦皇岛066004

出　　处：《华侨大学学报（自然科学版）》2013年第2期230-235,共6页Journal of Huaqiao University(Natural Science)

基　　金：国家自然科学基金资助项目(71071133)

摘　　要：把Q为正定矩阵或半正定矩阵推广到Q为亚正定矩阵,利用时滞投影神经网络模型和李亚普诺夫函数的特性,给出判断这种特殊二次优化最优解的充分条件.最后通过数值仿真说明了该网络的有效性.In this paper, we studied the stability of the optimal solution of a new quadratic programming problem which is the promotion of the convex quadratic programming. We give the sufficient condition to determine the stability of the equilibrium point and propose a delayed projection neural network to solve this quadratic programming. By constructing a suitable Lyapunov function, the proposed neural network is proved to be globally stable. Simulation results with some applications show the efficient of the proposed neural network.

关 键 词：亚正定矩阵 全局稳定 时滞投影 李雅普诺夫函数 最优解 

分 类 号：O175[理学—数学]